Method and apparatus for obtaining cerebral perfusion parameter images

Abstract

The application discloses a brain perfusion parameter image acquisition method and device, and relates to the technical field of perfusion imaging. A medical imaging device can acquire a plurality of perfusion scan images of a brain, acquire at least two target perfusion scan images from the plurality of perfusion scan images, process the at least two target perfusion scan images through a perfusion parameter map generation model, and obtain a brain perfusion parameter image. The at least two target perfusion scan images correspond to a plurality of key periods one by one. The perfusion parameter map generation model is trained by using a plurality of sample sets, each sample set includes a plurality of sample scan images corresponding to the plurality of key periods one by one and a sample perfusion parameter image. It can be seen that the method provided in the application can acquire the brain perfusion parameter image by processing the at least two target perfusion scan images through the perfusion parameter map generation model, and therefore, the method enriches the acquisition mode of the brain perfusion parameter image.

Description

Technical Field

[0001] This application relates to the field of perfusion imaging technology, and in particular to a method and apparatus for acquiring brain perfusion parameter images. Background Technology

[0002] In the field of brain perfusion imaging, medical imaging equipment can perform perfusion scans on the brain of the imaging subject, acquire multiple consecutive perfusion scan images, and extract the time-density curve (TDC) corresponding to each pixel in the perfusion scan image based on these multiple perfusion scan images.

[0003] Based on this, medical imaging equipment generates brain perfusion parameter images by quantitatively analyzing and fitting the time-density curves of each pixel, providing objective and accurate imaging evidence for clinical assessment of the brain blood flow perfusion status of the imaging subject and identification of abnormal perfusion areas (such as ischemia, hyperperfusion, etc.).

[0004] However, the current methods for obtaining brain perfusion parameter images based on perfusion scan images are still relatively limited. Summary of the Invention

[0005] This application provides a method and apparatus for acquiring brain perfusion parameter images, which can solve the problem that the methods for acquiring brain perfusion parameter images based on perfusion scan images in related technologies are relatively limited. The technical solution is as follows: On the one hand, a method for obtaining brain perfusion parameter images is provided, the method comprising: Obtain brain perfusion scan sequences, which include multiple perfusion scan images; At least two target perfusion scan images are selected from the brain perfusion scan sequence, wherein the at least two target perfusion scan images correspond one-to-one with multiple critical periods; The brain perfusion parameter image is obtained by processing the at least two target perfusion scan images using a perfusion parameter map generation model. The perfusion parameter map generation model is trained using multiple sample sets, each of which includes multiple sample scan images corresponding one-to-one with the multiple critical periods, and sample perfusion parameter images.

[0006] Optionally, the artifacts in each of the target perfusion scan images are less than a preset threshold, and the degree of motion offset is less than a degree threshold.

[0007] Optionally, at least two target perfusion scan images are selected from the brain perfusion scan sequence, including: A first perfusion scan image is identified from the brain perfusion scan sequence, wherein the motion offset of the first perfusion scan image is greater than or equal to the degree threshold, and / or the artifacts are greater than or equal to the preset threshold; The first perfusion scan image is removed from the brain perfusion scan sequence to obtain multiple second perfusion scan images; For each critical phase, the second perfusion scan image with the smallest time interval between the phase and the critical phase among the plurality of second perfusion scan images is determined as the target perfusion scan image corresponding to the critical phase.

[0008] Optionally, the second perfusion scan image with the smallest time interval between the phase and the critical phase among the plurality of second perfusion scan images is determined as the target perfusion scan image corresponding to the critical phase, including: Based on the brain perfusion scan sequence, the phase values ​​of each of the key phases are determined; The second perfusion scan image with the smallest absolute value of the difference between the phase value and the phase value of the key phase among the plurality of second perfusion scan images is determined as the second perfusion scan image with the smallest time interval between the phase and the key phase.

[0009] Optionally, based on the brain perfusion scan sequence, the phase values ​​of each of the key phases are determined, including: Based on the perfusion scan images other than the first perfusion scan image in the brain perfusion scan sequence, each of the first perfusion scan images is updated, wherein the motion offset of each of the updated first perfusion scan images is less than the degree threshold, and the artifacts are less than the preset threshold. Based on perfusion scan images other than the first perfusion scan image, and the updated first perfusion scan images, arterial inflow function curves and venous outflow function curves are obtained; Based on the arterial inflow function curve and the venous outflow function curve, the phase values ​​of the multiple key phases are determined.

[0010] Optionally, the at least two target perfusion scan images are processed using a perfusion parameter map generation model to obtain brain perfusion parameter images, including: By generating a perfusion parameter map model, the at least two target perfusion scan images and the time interval between the phases of each target perfusion scan image and the corresponding key phases are processed to obtain brain perfusion parameter images; Each of the sample sets further includes: the sample time interval between the phase of each sample scan image and the corresponding key phase.

[0011] Optionally, the motion shift degree of the first perfusion scan image is greater than or equal to the degree threshold; identifying the first perfusion scan image from the brain perfusion scan sequence includes: A reference scan image is identified from the brain perfusion scan sequence. The reference scan image has the highest global skull overlap. The global skull overlap is proportional to the sum of multiple skull overlaps corresponding to the reference scan image. The multiple skull overlaps include at least the overlap between the skull in the reference scan image and the skull in each perfusion scan image other than the reference scan image. Calculate the similarity between each perfusion scan image in the brain perfusion scan sequence, excluding the reference scan image, and the reference scan image; The perfusion scan images with a similarity less than the similarity threshold are identified as the first perfusion scan images.

[0012] Optionally, a reference scan image is identified from the brain perfusion scan sequence, including: For each perfusion scan image in the brain perfusion scan sequence, obtain multiple skull overlap values ​​corresponding to the perfusion scan image; The global cranial overlap corresponding to the perfusion scan image is determined based on the sum of multiple cranial overlap values ​​corresponding to the perfusion scan image. The perfusion scan image corresponding to the largest global skull overlap is determined as the reference scan image.

[0013] On the other hand, a device for acquiring brain perfusion parameter images is provided, the device comprising: The first acquisition module is used to acquire brain perfusion scan sequences, which include multiple perfusion scan images; A screening module is used to select at least two target perfusion scan images from the brain perfusion scan sequence, wherein the at least two target perfusion scan images correspond one-to-one with multiple critical phases; The second acquisition module is used to process the at least two target perfusion scan images through a perfusion parameter map generation model to obtain brain perfusion parameter images; The perfusion parameter map generation model is trained using multiple sample sets, each of which includes multiple sample scan images corresponding one-to-one with the multiple critical periods, and sample perfusion parameter images.

[0014] In another aspect, a medical imaging device is provided, the medical imaging device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the method for acquiring brain perfusion parameter images as described above.

[0015] In another aspect, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method for acquiring brain perfusion parameter images as described above.

[0016] In another aspect, a computer program product is provided, the computer program product comprising a computer program that, when executed by a processor, implements the method for acquiring brain perfusion parameter images as described above.

[0017] The beneficial effects of the technical solution provided in this application include at least the following: This application provides a method and apparatus for acquiring brain perfusion parameter images. The medical imaging equipment can acquire multiple perfusion scan images of the brain and extract at least two target perfusion scan images from these multiple images. These at least two target perfusion scan images are then processed by a perfusion parameter map generation model to obtain brain perfusion parameter images. The at least two target perfusion scan images correspond one-to-one with multiple critical periods. The perfusion parameter map generation model is trained using multiple sample sets, each sample set including multiple sample scan images corresponding one-to-one with multiple critical periods, and sample perfusion parameter images. Therefore, the method provided in this application can acquire brain perfusion parameter images by processing at least two target perfusion scan images using a perfusion parameter map generation model, thus enriching the methods for acquiring brain perfusion parameter images.

[0018] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0019] Figure 1 This is a flowchart of a method for obtaining brain perfusion parameter images provided in an embodiment of this application; Figure 2 This is a flowchart of another method for obtaining brain perfusion parameter images provided in an embodiment of this application; Figure 3 This is a schematic diagram of an AIF-VOF curve provided in an embodiment of this application; Figure 4 This is a schematic diagram of a sample scan image corresponding to a critical period and a sample perfusion parameter image provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a device for acquiring brain perfusion parameter images provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of a medical imaging device provided in an embodiment of this application. Detailed Implementation

[0020] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0021] This application provides a method for acquiring brain perfusion parameter images, applied to medical imaging equipment. This medical imaging equipment can be a computed tomography (CT) device or a magnetic resonance imaging (MRI) device. See also... Figure 1 The method includes: Step 101: Obtain brain perfusion scan sequences.

[0022] The brain perfusion scan sequence includes multiple perfusion scan images.

[0023] In this embodiment, the brain perfusion scan sequence may be pre-stored by a medical imaging device. Alternatively, the medical imaging device may perform a perfusion scan (such as a CT perfusion scan or an MR perfusion scan) on the brain of the imaging subject to obtain the brain perfusion scan sequence.

[0024] Optionally, at least one of the multiple perfusion scan images can be acquired during motion of the imaging object. That is, the imaging object moves during the acquisition of multiple perfusion scan images. Consequently, at least one perfusion scan image acquired will contain motion artifacts and / or exhibit motion shift.

[0025] Step 102: Select at least two target perfusion scan images from the brain perfusion scan sequence.

[0026] In this study, at least two target perfusion scan images correspond one-to-one with multiple key phases. These key phases may include: contrast agent initiation phase (also known as contrast agent initiation phase), arterial peak phase, venous peak phase, and late venous phase. The contrast agent initiation phase refers to the period when the contrast agent concentration-time curve in the artery transitions from a stable state to a clear increase, i.e., the period when the contrast agent begins to enter the artery. The arterial peak phase refers to the period when the contrast agent concentration in the artery reaches its highest point. The venous peak phase refers to the period when the contrast agent concentration in the vein reaches its highest point. The late venous phase refers to the period when the contrast agent concentration in the vein decreases to half of its peak value.

[0027] Understandably, artifacts in the target perfusion scan image can be less than a preset threshold, and the degree of motion offset can be less than a threshold. This ensures high accuracy of the obtained brain perfusion parameter images.

[0028] In this embodiment, the phase of the target perfusion scan image corresponding to the critical phase can be the critical phase itself. For example, the imaging object does not move during the acquisition of the brain perfusion scan sequence; or, the imaging object does not move during the acquisition of perfusion scan images of the critical phase. In these scenarios, the phase of the selected target perfusion scan image corresponding to the critical phase is the critical phase.

[0029] Alternatively, the phase of the target perfusion scan image corresponding to the critical phase should have the smallest time interval with that critical phase. For example, during the acquisition of brain perfusion scan sequences, the imaging object moves, and the phase of the movement happens to be the critical phase. Since the artifacts in the target perfusion scan image are required to be less than the artifact threshold, and the degree of motion offset is required to be less than the degree threshold, and in scenarios involving motion, it is necessary to select the perfusion scan image closest to the critical phase to determine the brain perfusion parameter image, in this scenario, the selected phase of the target perfusion scan image corresponding to the critical phase should have the smallest time interval with that critical phase. For example, the phase of the target perfusion scan image can be an adjacent phase to the critical phase.

[0030] Based on the above description, when the imaging object does not move during the acquisition of brain perfusion scan sequences, medical imaging equipment can directly identify the perfusion scan image with the critical phase as the target perfusion scan image.

[0031] In cases where the imaging object moves during the acquisition of a brain perfusion scan sequence, in one optional implementation, the medical imaging device can identify a first perfusion scan image from the brain perfusion scan sequence. The motion offset of the first perfusion scan image is greater than or equal to a threshold, and / or, artifacts are greater than or equal to a preset threshold. Subsequently, the medical imaging device can remove the first perfusion scan image from the brain perfusion scan sequence to obtain multiple second perfusion scan images. For each critical phase, the medical imaging device can identify the second perfusion scan image with the smallest time interval between the intermediate phase and the critical phase from the multiple second perfusion scan images as a target perfusion scan image corresponding to the critical phase, thus obtaining at least two target perfusion scan images.

[0032] In another alternative implementation, the medical imaging device pre-stores an image recognition model. The medical imaging device can input a brain perfusion scan sequence into this image recognition model to obtain at least two target perfusion scan images output by the image recognition model.

[0033] It should be understood that before inputting brain perfusion scan sequences into the image recognition model, medical imaging equipment can acquire multiple training sets and train the model on these sets to obtain the image recognition model. Each training set may include: a set of sample images and multiple key scan images selected from the sample image set. The sample image set includes multiple sample scan images. Each key scan image corresponds to a key phase. The time interval between the phase of each key scan image and its corresponding key phase is minimized, and the artifacts and motion shift of each sample key scan image are less than a preset threshold.

[0034] Step 103: Process at least two target perfusion scan images using the perfusion parameter map generation model to obtain brain perfusion parameter images.

[0035] The medical imaging equipment is pre-deployed with a perfusion parameter map generation model. After obtaining at least two target perfusion scan images, the medical imaging equipment can input these two target perfusion scan images into the perfusion parameter map generation model to obtain brain perfusion parameter images output by the perfusion parameter map generation model.

[0036] The number of brain perfusion parameter images can be one or more. Each brain perfusion parameter image corresponds one-to-one with a specific brain perfusion parameter. These parameters may include: cerebral blood volume (CBV), cerebral blood flow (CBF), mean transit time (MTT), time to peak (TTP), and time to maximum (Tmax) of the residual function.

[0037] The perfusion parameter map generation model was trained using multiple sample sets. Each sample set includes multiple sample scan images that correspond one-to-one with multiple critical periods, as well as sample perfusion parameter images.

[0038] In summary, this application provides a method for acquiring brain perfusion parameter images. A medical imaging device can acquire multiple perfusion scan images of the brain and extract at least two target perfusion scan images from these images. These at least two target perfusion scan images are then processed by a perfusion parameter map generation model to obtain the brain perfusion parameter image. Each of the at least two target perfusion scan images corresponds one-to-one with multiple critical periods. The perfusion parameter map generation model is trained using multiple sample sets, each sample set including multiple sample scan images corresponding one-to-one with multiple critical periods, and sample perfusion parameter images. Therefore, the method provided in this application can acquire brain perfusion parameter images by processing at least two target perfusion scan images using a perfusion parameter map generation model, thus enriching the methods for acquiring brain perfusion parameter images.

[0039] Taking the motion of the imaging object during the acquisition of a brain perfusion scan sequence, and the acquisition of at least two target perfusion scan images from multiple second perfusion scan images obtained after removing the first perfusion scan image, as an example, the method for acquiring brain perfusion parameter images provided in this application is illustrated. This method can be applied to medical imaging equipment. See also Figure 2 The method may include: Step 201: Obtain brain perfusion scan sequences.

[0040] The brain perfusion scan sequence comprises multiple perfusion scan images. The brain perfusion scan sequence can be pre-stored by the medical imaging equipment. Alternatively, the medical imaging equipment can scan the brain of the subject to obtain the brain perfusion scan sequence.

[0041] Step 202: Identify the first perfusion scan image from the brain perfusion scan sequence.

[0042] The number of first perfusion scan images can be one or more. The motion offset degree of each first perfusion scan image is greater than or equal to a threshold, and / or the artifacts are greater than or equal to a preset threshold. Therefore, the first perfusion scan image is an abnormal perfusion scan image.

[0043] Artifacts can include at least one of metallic artifacts and motion artifacts. An artifact percentage greater than or equal to a preset threshold can mean that the proportion of artifacts is greater than or equal to a preset percentage (i.e., the preset threshold). The artifact percentage can be the ratio of the area containing the artifact to the area of ​​the perfusion scan image.

[0044] Optionally, the multiple first perfusion scan images can include all perfusion scan images containing artifacts. In this case, the preset percentage should be set relatively small, such as 3%. This ensures higher accuracy of the subsequently generated brain perfusion parameter images.

[0045] Metal artifacts and motion artifacts have obvious artifact characteristics, while the degree of motion offset needs to be determined by comparing different perfusion scan image data. Taking a perfusion scan image containing artifacts as an example, the process of medical imaging equipment identifying the first perfusion scan image is illustrated below. This process may include: Step S1: Register multiple perfusion scan images in the brain perfusion scan sequence.

[0046] During the scanning process, slight movements of the subject (such as slight head movements, swallowing, respiratory micro-movements, or eye movements) can cause shifts in the position of brain tissue in different perfusion scan images. This shift can affect the recognition accuracy of the first perfusion scan image, thereby affecting the reliability of the acquired brain perfusion parameter images.

[0047] Based on this, medical imaging equipment can register multiple perfusion scan images to strictly align them, thereby eliminating positional deviations caused by slight movements and greatly improving the accuracy of the first perfusion scan image being identified.

[0048] Optionally, the medical imaging equipment can perform rigid registration on multiple perfusion scan images to align them. Rigid registration refers to performing rigid transformations such as rotation, translation, and scaling between different perfusion scan images to align them spatially. This preserves as much of the original shape of the perfusion scan images as possible and minimizes image distortion.

[0049] Specifically, medical imaging equipment can perform rigid registration on perfusion scan images to obtain a registration transformation matrix. This registration transformation matrix can include multiple registration parameters such as rotation angle and translation. Subsequently, the medical imaging equipment can use this registration transformation matrix to perform spatial transformation on the perfusion scan image to obtain a registered perfusion scan image.

[0050] Optionally, the registration parameters in a 3D scene can include translation in three coordinate axes and rotation angles in three axes, for a total of 6 independent degrees of freedom parameters.

[0051] It should be understood that during the spatial transformation of the perfusion scan image using the registration transformation matrix, if the coordinate points obtained by the transformation are not integers, the pixel value (such as gray value) of the current coordinate point can be estimated by interpolation, thereby achieving the registration of the perfusion scan image.

[0052] The interpolation method can be one of the following: bilinear interpolation and bicubic interpolation, such as bilinear interpolation.

[0053] Step S2: Identify perfusion scan images containing artifacts from multiple perfusion scan images.

[0054] In one possible embodiment, the medical imaging device can preprocess the registered perfusion scan images. Then, the medical imaging device can input the preprocessed perfusion scan images into a pre-deployed artifact recognition model to obtain the recognition result output by the artifact recognition model, and subsequently obtain the perfusion scan image containing artifacts based on the recognition result.

[0055] The identification results include: no artifacts, presence of motion artifacts, and presence of metal artifacts. Medical imaging equipment can identify perfusion scan images with motion artifacts and metal artifacts as perfusion scan images with artifacts.

[0056] Optionally, the artifact recognition model can be a deep residual network (ResNet) model, such as ResNet-34. ResNet-34 is a deep residual network consisting of 34 convolutional layers. Its core structure contains four layers composed of stacked residual blocks. Layer 1 contains 3 residual blocks, Layer 2 contains 4, Layer 3 contains 6, and Layer 4 contains 3. Each residual block is formed by two 3×3 convolutional layers (stretch 1, padding 1) and a shortcut across layers. Specifically, the first residual block of Layers 2, 3, and 4 uses a convolution with a stretch of 2 to halve the feature map size, and adjusts the number of channels using a 1×1 convolution to match the output of the main branch. The network ends with a global average pooling layer and a fully connected layer, ultimately outputting a 3-class classification result: no artifacts, presence of motion artifacts, and presence of metal artifacts.

[0057] In this embodiment, preprocessing may include spatial resolution normalization and grayscale normalization. Resolution normalization refers to: statistically analyzing the three-dimensional spatial spacing of voxels in all perfusion scan images of the brain perfusion scan sequence, obtaining the median of spacing in each of the X, Y, and Z dimensions, and uniformly normalizing the spacing of each dimension of all perfusion scan images to the median of spacing in the corresponding dimension. The median spacing refers to the median value of spacing. This spatial spacing refers to the actual distance between two adjacent voxels in the image in physical space. Assuming the position of the voxels before spatial resolution normalization is ( x , y , z Then the position of the spatially resolved normalized voxel can be ( , , )satisfy: Formula (1) In formula (1), Spacing of the X dimension of a voxel before spatial resolution normalization. The median of the spacing in the X dimension. Spacing of the Y dimension of a voxel before spatial resolution normalization. This represents the median of the spacing along the Y dimension. Spacing of voxels in the Z direction before spatial resolution normalization. The median of the spacing along the Z dimension.

[0058] The process of grayscale normalization for each perfusion scan image may include: statistically analyzing the grayscale values ​​of all pixels located within brain tissue in all perfusion scan images, and calculating the mean and standard deviation of the grayscale values ​​of all pixels; obtaining the normalized grayscale value of each pixel using the following formula. : Formula (2) In formula (2), The grayscale value before normalization. The average grayscale value of all pixels. is the standard deviation of the grayscale values ​​of all pixels.

[0059] Step S3: Identify perfusion scan images from multiple perfusion scan images whose motion offset is greater than or equal to the degree threshold.

[0060] Medical imaging equipment can first identify a reference scan image from multiple perfusion scan images. Subsequently, the medical imaging equipment can calculate the similarity between each perfusion scan image (excluding the reference scan image) and the reference scan image, and identify perfusion scan images with similarity scores less than a similarity threshold as perfusion scan images with motion offset degrees greater than or equal to a threshold.

[0061] Among these, the reference scan image corresponds to the highest global skull overlap. This global skull overlap can be proportional to the sum of multiple skull overlap values ​​corresponding to the reference scan image, such as this sum. The multiple skull overlap values ​​include at least the overlap between the skull in the reference scan image and the skull in each of the other perfusion scan images. That is, the reference scan image is the most stable perfusion scan image among the multiple perfusion scan images.

[0062] For example, multiple skull overlap measurements can include: the overlap between the skull in the reference scan image and the skull in each (i.e., each) perfusion scan image in the brain perfusion scan sequence. Alternatively, multiple skull overlap measurements can include: the overlap between the skull in the reference scan image and the skull in each perfusion scan image in the brain perfusion scan sequence other than the reference scan image. This overlap can be characterized using cross-union ratio (CUNR).

[0063] Optionally, this similarity can be represented using mutual information, in which case the similarity threshold is the mutual information threshold. (Refer to scanned image) A With each perfusion scan image B mutual information I ( A , B It can satisfy: Formula (3) In formula (3), p ( a , b (Reference scan image) A With perfusion scan images B The joint probability function. p ( a (Reference scan image) A marginal probability function, p ( b ) is a perfusion scan image B The marginal probability function.

[0064] The following is an exemplary description of the process for obtaining the reference scanned image: For each perfusion scan image in a brain perfusion scan sequence, the medical imaging device can acquire multiple cranial overlap values ​​corresponding to the perfusion scan image, and determine the global cranial overlap value corresponding to that perfusion scan image based on the sum of these multiple cranial overlap values. This global cranial overlap value can be positively correlated with the sum value, such as being equal to the sum. Subsequently, the medical imaging device can compare the global cranial overlap values ​​corresponding to multiple perfusion scan images to obtain the maximum global cranial overlap value, and determine the perfusion scan image corresponding to the maximum global cranial overlap value as the reference scan image.

[0065] In this embodiment, for each perfusion scan image, the medical imaging device can calculate the degree of overlap between the skull in that perfusion scan image and the skull in each perfusion scan image in the brain perfusion scan sequence, thus obtaining multiple corresponding skull overlaps. At this point, the total number of these multiple skull overlaps is equal to the total number of all perfusion scan images included in the brain perfusion scan sequence.

[0066] Alternatively, the medical imaging equipment can calculate the degree of overlap between the skull in the perfusion scan image and the skull in every other perfusion scan image in the brain perfusion scan sequence, thus obtaining multiple skull overlap values. In this case, the total number of skull overlap values ​​is equal to the difference between the total number of all perfusion scan images included in the brain perfusion scan sequence and 1. This reduces the computational resources of the medical imaging equipment and improves the efficiency of acquiring reference scan images.

[0067] Step S4: The perfusion scan images containing artifacts and the perfusion scan images with motion offset greater than or equal to the degree threshold are all identified as the first perfusion scan images.

[0068] After identifying perfusion scan images with artifacts and perfusion scan images with motion offset greater than or equal to a threshold from multiple perfusion scan images, the medical imaging equipment can determine each of the perfusion scan images with artifacts and perfusion scan images with motion offset greater than or equal to the threshold as a first perfusion scan image.

[0069] Step 203: Remove the first perfusion scan image from the brain perfusion scan sequence to obtain multiple second perfusion scan images.

[0070] After the medical imaging equipment identifies the first perfusion scan image, it can be removed from the brain perfusion scan sequence to obtain multiple second perfusion scan images. These multiple second perfusion scan images are normal perfusion scan images with artifacts less than a preset threshold and motion deviation less than a threshold.

[0071] Step 204: For each critical phase, the second perfusion scan image with the smallest time interval between the phase and the critical phase among the multiple second perfusion scan images is determined as the target perfusion scan image corresponding to the critical phase.

[0072] Among these, the motion shift in the perfusion scan images of each target is less than a threshold, and the artifacts are less than a preset threshold. Several key phases include: contrast agent initiation, arterial peak concentration, venous peak concentration, and late venous phase.

[0073] Understandably, for each key phase in the contrast agent initiation phase, arterial peak phase, and venous peak phase, if the medical imaging equipment determines that among multiple second perfusion scan images, the number of second perfusion scan images with the smallest time interval between the phase and the key phase is two, then the second perfusion scan image with a phase later than the key phase can be identified as the target perfusion scan image corresponding to the key phase.

[0074] For late venous disease, if the medical imaging equipment determines that among multiple second perfusion scan images, the number of second perfusion scan images with the smallest time interval between the critical phase and the critical phase is two, then the second perfusion scan image with a phase earlier than the late venous disease can be identified as the target perfusion scan image corresponding to the late venous disease.

[0075] In this embodiment, the medical imaging device can determine the phase value of each critical phase based on the brain perfusion scan sequence. For each critical phase, the medical imaging device can determine the second perfusion scan image with the smallest absolute value of the difference between the phase value and the phase value of the critical phase from among multiple second perfusion scan images, as the second perfusion scan image with the smallest time interval between the phase and the critical phase.

[0076] Here, the absolute value of the difference refers to the absolute value of the difference between the phase values. The phase value of the key phase, as well as each phase in the phase of the second perfusion scan image, can be represented by a moment within the time period of that phase.

[0077] The following is a brief explanation of the process for determining the phase values ​​of each key phase based on a brain perfusion scan sequence: The medical imaging equipment can update each of the first perfusion scan images in the brain perfusion scan sequence based on the perfusion scan images excluding the first perfusion scan image. The motion offset of each updated first perfusion scan image is less than a threshold, and the artifacts are less than a preset threshold. Then, based on the perfusion scan images excluding the first perfusion scan image in the brain perfusion scan sequence, and the updated first perfusion scan images, the medical imaging equipment can acquire the arterial inflow function (AIF) curve and the venous outflow function (VOF) curve, and determine the phase values ​​of multiple key phases based on these curves.

[0078] The arterial inflow function curve is the core curve reflecting the concentration-time change of the contrast agent before it enters the tissue via the artery. The venous outflow function curve reflects the concentration-time change of the contrast agent after it has been metabolized by the tissue and flows out via the vein. Furthermore, both the arterial inflow and venous outflow function curves are composed of discrete points, each corresponding to a perfusion scan image, i.e., a phase. However, in actual display and use, these curves are generally processed using mathematical fitting or interpolation to obtain a smoother curve that more closely approximates the real, continuous hemodynamic process.

[0079] Optionally, the medical imaging device can use interpolation to update each first perfusion scan image based on perfusion scan images other than the first perfusion scan image in the brain perfusion scan sequence. Specifically, for each first perfusion scan image, the medical imaging device can use interpolation to process the two second perfusion scan images that are closest in phase distance before the first perfusion scan image and closest in phase distance after the second perfusion scan image, to obtain the updated first perfusion scan image.

[0080] The following example illustrates how medical imaging equipment determines the phase values ​​of multiple key phases based on arterial inflow function curves and venous outflow function curves, using a single moment within the time period of the phase as an example: Both the arterial and venous peak contrast periods are characterized by the highest contrast agent density. The maximum density of the arterial inflow function curve represents the peak contrast agent concentration in the artery, while the maximum density of the venous outflow function curve represents the peak contrast agent concentration in the vein. Therefore, medical imaging equipment can determine the phase value of the arterial peak contrast period based on the highest point (peak value) of the arterial inflow function curve, and similarly, the phase value of the arterial peak contrast period based on the highest point of the venous outflow function curve.

[0081] The late venous phase occurs at the midpoint of the venous outflow function curve. Therefore, medical imaging equipment can define the late venous phase as the point in the venous outflow function curve where the contrast agent concentration drops to half of its peak value after the peak venous phase.

[0082] The contrast agent initiation period is located at the starting point of the main peak of the arterial inflow function curve. Therefore, the process of acquiring the phase value of the contrast agent initiation period using medical imaging equipment can include: acquiring multiple target discrete points in the arterial inflow function curve within a target time interval, where the lower limit of the target time interval is the time corresponding to the starting point of the arterial inflow function curve, and the upper limit is the time corresponding to the highest point; performing dynamic curve fitting on these multiple target discrete points using a gamma variable model to obtain a fitted curve; solving for the characteristic time corresponding to the global positive maximum value of the second derivative of the fitted curve; and determining the time closest to this characteristic time among the times corresponding to all discrete points constituting the arterial inflow function curve (i.e., the original sampling time sequence) as the phase value of the contrast agent initiation period. Here, the original sampling time sequence is the acquisition time sequence of the brain perfusion image sequence.

[0083] Optionally, before acquiring multiple target discrete points in the arterial inflow function curve, the medical imaging equipment can also perform baseline correction and smoothing / denoising on the arterial inflow function curve. This can largely eliminate interference signals, thereby improving the accuracy of determining the phase value at the onset of the contrast agent.

[0084] For example, see Figure 3 , Figure 3 A schematic diagram of an arterial inflow function curve and a venous outflow function curve is shown. Figure 3 The arterial inflow function curve and venous outflow function curve shown are both mathematically fitted function curves. At this point, the medical imaging equipment can determine the time corresponding to point A as the phase value of the contrast agent initiation phase; the time corresponding to point B as the phase value of the arterial peak phase; the time corresponding to point C as the phase value of the venous peak phase; and the time corresponding to point D as the phase value of the late venous phase.

[0085] In this embodiment, the medical imaging device can select arterial input points (AIPs) and venous output points (VOPs) manually or automatically based on perfusion scan images other than the first perfusion scan image in the brain perfusion scan sequence, as well as the updated first perfusion scan images. It then extracts the acquisition time and grayscale value of each input point in each perfusion scan image. Subsequently, the medical imaging device can obtain an arterial inflow function curve based on the acquisition time and grayscale value of the arterial input point in each perfusion scan image, and can obtain a venous outflow function curve based on the acquisition time and grayscale value of the venous input point in each perfusion scan image.

[0086] Step 205: Process at least two target perfusion scan images using the perfusion parameter map generation model to obtain brain perfusion parameter images.

[0087] The perfusion parameter map generation model is trained using multiple sample sets. Each sample set includes multiple sample scan images corresponding one-to-one with multiple critical periods, and sample perfusion parameter images. The sample perfusion parameter images are determined based on a set of sample images collected during non-motor periods. This results in the brain perfusion parameter images generated by the perfusion parameter map generation model exhibiting extremely high consistency with real brain perfusion parameter images.

[0088] Multiple perfusion parameters can include: CBV, CBF, MTT, time-to-peak (TTP) images, and time-to-maximum (Tmax) images, etc. Correspondingly, the number of brain perfusion parameter images can be one or more, such as multiple images. Multiple brain perfusion parameter images can include: CBV images, CBF images, MTT images, TTP images, and Tmax images, etc. That is, medical imaging equipment can acquire a brain perfusion map based on at least two target perfusion scan images.

[0089] In this embodiment, the medical imaging device is pre-deployed with a perfusion parameter map generation model. The medical imaging device can input at least two target perfusion scan images into the perfusion parameter map generation model to obtain a CBV image, CBF image, TTP image, and Tmax image output by the model. Then, based on the relationship between MTT, CBV, and CBF: MTT = CBV / CBF, the medical imaging device can obtain the MTT image based on the CBV image and the CBF image.

[0090] Optionally, each sample set may further include: the sample time interval between the phase of each sample scan image and the corresponding key phase. This makes the trained perfusion parameter map generation model applicable not only to acquiring brain perfusion parameter images in non-motion scenarios but also in motion scenarios. Thus, the applicability of the method provided in this application embodiment is improved.

[0091] At this time, the medical imaging device may perform step 205 by: generating a model through perfusion parameter maps, processing at least two target perfusion scan images, and the time interval between the phases of each target perfusion scan image and the corresponding key phases, to obtain brain perfusion parameter images.

[0092] The training process of the perfusion parameter map generation model is illustrated below, taking each sample set, including sample time intervals, as an example: Medical imaging equipment can acquire a set of m normal, motion-free, and artifact-free sample images, each set of sample images comprising multiple sample scan images. For each sample image set, such as... Figure 4 As shown, medical imaging equipment can obtain multiple sample scan images corresponding to multiple key periods in a sample set based on this sample image set, as well as the sample time interval between the phase of each sample scan image and the corresponding key period phase. These are denoted as (data1, time1), (data2, time2), (data3, time3), and (data4, time4). Here, data represents the sample scan image, and time represents the sample time interval. A positive time indicates an earlier key period phase, and a negative time indicates a later key period phase. In this case, time1 = time2 = time3 = time4 = 0.

[0093] Furthermore, the medical imaging equipment can randomly acquire sample scan images from the sample image set whose phases are near the aforementioned critical phases, and obtain the sample time interval between the phase and the corresponding critical phase, to obtain multiple sample scan images corresponding to multiple critical phases in another sample set, as well as the sample time interval between the phase of each sample scan image and the corresponding critical phase. These are also denoted as (data1, time1), (data2, time2), (data3, time3), and (data4, time4). By repeatedly executing the above random acquisition process, n sample sets can be obtained. Here, n can be used as a variable parameter to adjust the diversity of the sample sets. Thus, m×n sample sets can be formed.

[0094] For each sample image set, the medical imaging equipment employs traditional perfusion parameter calculation methods, such as the maximum slope method, partial deconvolution method, singular value decomposition (SVD) method, and block-circulant deconvolution method, to obtain... Figure 4 The sample perfusion parameter images shown serve as labels for sample sets derived from this set of sample images. It should be understood that multiple sample sets derived from this set of sample images may have the same label. For example... Figure 4 As shown, multiple sample perfusion parameter images may include: CBF sample image, CBV sample image, TTP sample image and Tmax sample image.

[0095] For an m×n sample set, the medical imaging equipment can first perform standardization and data augmentation processing on the m×n sample set, such as bone removal, grayscale normalization, and spacing normalization. Then, the medical imaging equipment can train a model on the processed m×n sample set to obtain a perfusion parameter map generation model.

[0096] During the training of the perfusion parameter map generation model, for each sample set, the scanned images of each sample in the sample set, as well as the sample time interval between the phase of the scanned image and the corresponding key phase, are fused through channel splicing and encoding embedding.

[0097] Before training the perfusion parameter map generation model, a basic generative model, i.e., a baseline system, is first constructed using standard VAEs, GANs, etc. Then, the perfusion parameter map generation model can be trained based on the baseline system. During training, a diffusion model is gradually introduced to improve the stability and detail fidelity of the generated results. Furthermore, a combination of loss functions, including adversarial loss, perceptual loss, and structural similarity (SSIM) constraints, can be used for model training to improve the reliability of the perfusion parameter map generation model, thereby ensuring the clinical credibility of the results generated using this model.

[0098] Optionally, after obtaining brain perfusion parameter images, medical imaging equipment can also obtain lesion segmentation results based on these images. For example, a threshold segmentation method can be used to process the brain perfusion parameter images to obtain lesion segmentation results.

[0099] Furthermore, during the training of the perfusion parameter map generation model, the sample segmentation results obtained based on the sample perfusion parameter images can be used as a supervision signal to calculate the loss function for model training, thereby ensuring the performance of lesion segmentation in brain perfusion parameter images generated by the perfusion parameter map generation model.

[0100] In the field of brain perfusion imaging, medical imaging equipment can perform perfusion scans on the brain of the imaging subject, acquiring multiple consecutive perfusion scan images. Based on these multiple perfusion scan images, the temporal density curve of each pixel is obtained. Subsequently, the medical imaging equipment can acquire brain perfusion parameter images based on the temporal density curves of each pixel. It is evident that the current methods for acquiring brain perfusion parameter images are relatively limited.

[0101] Furthermore, movement during the scanning process can cause significant motion artifacts in the perfusion scan images, making post-processing analyses such as brain perfusion parameter calculations impossible. Therefore, it is necessary to keep the subject's body still during the scan. However, in actual scanning, the subject may involuntarily move due to their condition, even after immobilization. This can render the perfusion scan images unsuitable for post-processing calculations of brain perfusion parameters, affecting diagnosis and treatment, and wasting valuable time and scanning resources.

[0102] For motion-related events during the aforementioned scanning process, a second rescan is often performed to obtain new, artifact-free perfusion scan images, which are then used to acquire brain perfusion parameter images. However, acquiring brain perfusion parameter images through a second rescan is inefficient and wastes medical resources.

[0103] Alternatively, the perfusion scan images with severe motion artifacts can be removed from the multiple perfusion scan images obtained, and the time density curves can be obtained based on the remaining perfusion scan images to obtain brain perfusion parameter images.

[0104] While deleting perfusion scans with severe motion artifacts and then using the remaining scans to obtain brain perfusion parameter images avoids interference from motion artifacts, the resulting temporal density curves for calculating brain perfusion parameters are incomplete. This leads to discrepancies between the brain perfusion parameter images calculated from the remaining scans and those calculated from the complete temporal density curves, and the degree of discrepancy increases with the number of deleted perfusion scans. Furthermore, in actual scanning, deleting 2 to 3 perfusion scans from the arrival of the contrast agent to the end of imaging often results in significant deviations in the calculated brain perfusion parameter images, potentially causing misdiagnosis. Therefore, the accuracy of brain perfusion parameter images obtained using this method is relatively low.

[0105] This application provides a novel method for acquiring brain perfusion parameter images. This method involves acquiring multiple perfusion scan images of the brain, extracting at least two target perfusion scan images from these multiple images, and then processing these at least two target perfusion scan images using a perfusion parameter map generation model to obtain the brain perfusion parameter image. This enriches the methods for acquiring brain perfusion parameter images.

[0106] Since brain perfusion parameter images can be obtained by processing at least two target perfusion scan images through a trained perfusion parameter map generation model, for scenarios where motion occurs during perfusion scanning, the method provided in this application embodiment only needs to find the target perfusion scan image with the smallest artifacts, motion offset, and time interval between the phase and the corresponding key phase from multiple perfusion scan images to reconstruct a real brain perfusion parameter image.

[0107] Compared to secondary rescanning, this method effectively improves the efficiency of acquiring brain perfusion parameter images and can significantly save medical resources. Compared to the method of deleting perfusion scans with severe motion artifacts and then acquiring brain perfusion parameter images based on the remaining perfusion scans, this method ensures higher accuracy and consistency with real brain perfusion parameter images because the artifacts in each target perfusion scan image are less than a preset threshold and the degree of motion offset is less than a threshold.

[0108] As can be seen, the method provided in this application can largely avoid motion interference in the acquisition of perfusion parameter images. Therefore, even if the imaging object moves during perfusion scanning, this method can still reconstruct a true perfusion parameter image. This significantly increases the tolerance of perfusion scanning to motion and improves the success rate of post-processing calculations of perfusion parameter images.

[0109] It is understood that the order of steps in the method for acquiring brain perfusion parameter images provided in the embodiments of this application can be appropriately adjusted, and steps can also be added or removed as appropriate. Any variations that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the protection scope of this application, and therefore will not be elaborated further.

[0110] In summary, this application provides a method for acquiring brain perfusion parameter images. A medical imaging device can acquire multiple perfusion scan images of the brain and extract at least two target perfusion scan images from these images. These at least two target perfusion scan images are then processed by a perfusion parameter map generation model to obtain the brain perfusion parameter image. Each of the at least two target perfusion scan images corresponds one-to-one with multiple critical periods. The perfusion parameter map generation model is trained using multiple sample sets, each sample set including multiple sample scan images corresponding one-to-one with multiple critical periods, and sample perfusion parameter images. Therefore, the method provided in this application can acquire brain perfusion parameter images by processing at least two target perfusion scan images using a perfusion parameter map generation model, thus enriching the methods for acquiring brain perfusion parameter images.

[0111] This application provides an apparatus for acquiring brain perfusion parameter images, which can perform the brain perfusion parameter image acquisition method provided in the above-described method embodiments. See also Figure 5 The device 300 includes: The first acquisition module 301 is used to acquire brain perfusion scan sequences, which include multiple perfusion scan images; The screening module 302 is used to select at least two target perfusion scan images from the brain perfusion scan sequence, wherein the at least two target perfusion scan images correspond one-to-one with multiple critical phases; The second acquisition module 303 is used to process at least two target perfusion scan images through a perfusion parameter map generation model to obtain brain perfusion parameter images. The perfusion parameter map generation model is trained using multiple sample sets, each of which includes multiple sample scan images corresponding to multiple critical periods, as well as sample perfusion parameter images.

[0112] Optionally, the artifacts in the perfusion scan images of each target are less than a preset threshold, and the degree of motion offset is less than a degree threshold.

[0113] Optionally, the filtering module 302 can be used for: The first perfusion scan image is identified from the brain perfusion scan sequence. The degree of motion offset of the first perfusion scan image is greater than or equal to a degree threshold, and / or the artifacts are greater than or equal to a preset threshold. The first perfusion scan image was removed from the brain perfusion scan sequence to obtain multiple second perfusion scan images; For each critical phase, the second perfusion scan image with the smallest time interval between the phase and the critical phase among multiple second perfusion scan images is determined as the target perfusion scan image corresponding to the critical phase.

[0114] Optionally, the process by which the screening module 302 determines the second perfusion scan image with the smallest time interval between the critical phase and the key phase from among multiple second perfusion scan images may include: Based on brain perfusion scan sequences, the phase values ​​of each key phase were determined; Among multiple second perfusion scan images, the second perfusion scan image with the smallest absolute value of the difference between the phase value and the phase value of the key phase is determined as the second perfusion scan image with the smallest time interval between the phase and the key phase.

[0115] Optionally, the process by which the screening module 302 determines the phase values ​​of each key phase based on the brain perfusion scan sequence may include: Based on the perfusion scan images other than the first perfusion scan image in the brain perfusion scan sequence, each first perfusion scan image is updated, wherein the motion offset of each updated first perfusion scan image is less than the degree threshold and the artifacts are less than the preset threshold. Based on the perfusion scan images other than the first perfusion scan image, and the updated first perfusion scan images, the arterial inflow function curve and the venous outflow function curve are obtained; Based on the arterial inflow function curve and the venous outflow function curve, the phase values ​​of multiple key phases are determined.

[0116] Optionally, the second acquisition module 303 can be used for: By generating a perfusion parameter map model, at least two target perfusion scan images and the time interval between the phases of each target perfusion scan image and the corresponding key phases are processed to obtain brain perfusion parameter images. Each sample set also includes the sample time interval between the phase of each sample scan image and the corresponding key phase.

[0117] Optionally, the motion offset of the first perfusion scan image is greater than or equal to a threshold. The filtering module 302 can be used to: A reference scan image is identified from the brain perfusion scan sequence. The reference scan image has the highest global skull overlap. This global skull overlap is proportional to the sum of multiple skull overlaps corresponding to the reference scan image. The multiple skull overlaps include at least the overlap between the skull in the reference scan image and the skull in each perfusion scan image other than the reference scan image. Calculate the similarity between each perfusion scan image (excluding the reference scan image) and the reference scan image in the brain perfusion scan sequence; Perfusion scan images with a similarity less than the similarity threshold are identified as the first perfusion scan images.

[0118] Optionally, the process by which the screening module 302 identifies the reference scan image from the brain perfusion scan sequence may include: For each perfusion scan image in the brain perfusion scan sequence, obtain multiple skull overlap values ​​corresponding to that perfusion scan image; The global cranial overlap corresponding to the perfusion scan image is determined based on the sum of multiple cranial overlap values ​​corresponding to the perfusion scan image. The perfusion scan image corresponding to the largest global skull overlap was selected as the reference scan image.

[0119] In summary, this application provides a device for acquiring brain perfusion parameter images. This device can acquire multiple perfusion scan images of the brain and extract at least two target perfusion scan images from these multiple images. The at least two target perfusion scan images are then processed by a perfusion parameter map generation model to obtain a brain perfusion parameter image. Each of the at least two target perfusion scan images corresponds one-to-one with multiple critical periods. The perfusion parameter map generation model is trained using multiple sample sets, each sample set including multiple sample scan images corresponding one-to-one with multiple critical periods, and sample perfusion parameter images. Therefore, the device provided in this application can acquire brain perfusion parameter images by processing at least two target perfusion scan images using a perfusion parameter map generation model, thus enriching the methods for acquiring brain perfusion parameter images.

[0120] This application also provides a medical imaging device, see [link to relevant documentation]. Figure 6 The medical imaging device 400 may include a processor 401 and a memory 403. The processor 401 and the memory 403 are connected, for example, via a bus 402. Optionally, the controller 400 may also include a transceiver 404. It should be noted that in practical applications, the transceiver 404 is not limited to one, and the structure of the controller 400 does not constitute a limitation on the embodiments of this application.

[0121] Processor 401 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 401 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0122] Bus 402 may include a pathway for transmitting information between the aforementioned components. Bus 402 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 402 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0123] The memory 403 stores a computer program corresponding to the brain perfusion parameter image acquisition method provided in the above embodiments of this application. This computer program is controlled and executed by the processor 401. The processor 401 executes the computer program stored in the memory 403 to implement the content shown in the aforementioned method embodiments.

[0124] This application provides a computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the method for acquiring brain perfusion parameter images as provided in the above-described method embodiments. For example, Figure 1 or Figure 2 The method shown.

[0125] This application provides a computer program product, which includes a computer program that, when executed by a processor, implements the method for acquiring brain perfusion parameter images as provided in the above-described method embodiments. For example, Figure 1 or Figure 2 The method shown.

[0126] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0127] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0128] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0129] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0130] In this application, unless otherwise expressly specified and limited, the terms "installation," "connection," "joining," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise expressly limited. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0131] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for acquiring brain perfusion parameter images, characterized in that, The method includes: Obtain brain perfusion scan sequences, which include multiple perfusion scan images; At least two target perfusion scan images are selected from the brain perfusion scan sequence, wherein the at least two target perfusion scan images correspond one-to-one with multiple critical periods; The brain perfusion parameter image is obtained by processing the at least two target perfusion scan images using a perfusion parameter map generation model. The perfusion parameter map generation model is trained using multiple sample sets, each of which includes multiple sample scan images corresponding one-to-one with the multiple critical periods, and sample perfusion parameter images.

2. The method according to claim 1, characterized in that, The artifacts in each of the target perfusion scan images are less than a preset threshold, and the degree of motion offset is less than a degree threshold.

3. The method according to claim 2, characterized in that, Select at least two target perfusion scan images from the brain perfusion scan sequence, including: A first perfusion scan image is identified from the brain perfusion scan sequence, wherein the motion offset of the first perfusion scan image is greater than or equal to the degree threshold, and / or the artifacts are greater than or equal to the preset threshold; The first perfusion scan image is removed from the brain perfusion scan sequence to obtain multiple second perfusion scan images; For each critical phase, the second perfusion scan image with the smallest time interval between the phase and the critical phase among the plurality of second perfusion scan images is determined as the target perfusion scan image corresponding to the critical phase.

4. The method according to claim 3, characterized in that, Among the plurality of second perfusion scan images, the second perfusion scan image with the smallest time interval between the phase and the key phase is determined as the target perfusion scan image corresponding to the key phase, including: Based on the brain perfusion scan sequence, the phase values ​​of each of the key phases are determined; The second perfusion scan image with the smallest absolute value of the difference between the phase value and the phase value of the key phase among the plurality of second perfusion scan images is determined as the second perfusion scan image with the smallest time interval between the phase and the key phase.

5. The method according to claim 4, characterized in that, Based on the brain perfusion scan sequence, the phase values ​​of each of the key phases are determined, including: Based on the perfusion scan images other than the first perfusion scan image in the brain perfusion scan sequence, each of the first perfusion scan images is updated, wherein the motion offset of each of the updated first perfusion scan images is less than the degree threshold, and the artifacts are less than the preset threshold. Based on perfusion scan images other than the first perfusion scan image, and the updated first perfusion scan images, arterial inflow function curves and venous outflow function curves are obtained; Based on the arterial inflow function curve and the venous outflow function curve, the phase values ​​of the multiple key phases are determined.

6. The method according to claim 3, characterized in that, The brain perfusion parameter images are obtained by processing the at least two target perfusion scan images using a perfusion parameter mapping generation model, including: By generating a perfusion parameter map model, the at least two target perfusion scan images and the time interval between the phases of each target perfusion scan image and the corresponding key phases are processed to obtain brain perfusion parameter images; Each of the sample sets further includes: the sample time interval between the phase of each sample scan image and the corresponding key phase.

7. The method according to any one of claims 3 to 6, characterized in that, The degree of motion offset in the first perfusion scan image is greater than or equal to the degree threshold; The first perfusion scan image is identified from the brain perfusion scan sequence, including: A reference scan image is identified from the brain perfusion scan sequence. The reference scan image has the highest global skull overlap. The global skull overlap is proportional to the sum of multiple skull overlaps corresponding to the reference scan image. The multiple skull overlaps include at least the overlap between the skull in the reference scan image and the skull in each perfusion scan image other than the reference scan image. Calculate the similarity between each perfusion scan image in the brain perfusion scan sequence, excluding the reference scan image, and the reference scan image; The perfusion scan images with a similarity less than the similarity threshold are identified as the first perfusion scan images.

8. The method according to claim 7, characterized in that, A reference scan image is identified from the brain perfusion scan sequence, including: For each perfusion scan image in the brain perfusion scan sequence, obtain multiple skull overlap values ​​corresponding to the perfusion scan image; The global cranial overlap corresponding to the perfusion scan image is determined based on the sum of multiple cranial overlap values ​​corresponding to the perfusion scan image. The perfusion scan image corresponding to the largest global skull overlap is determined as the reference scan image.

9. A device for acquiring brain perfusion parameter images, characterized in that, The device includes: The first acquisition module is used to acquire brain perfusion scan sequences, which include multiple perfusion scan images; A screening module is used to select at least two target perfusion scan images from the brain perfusion scan sequence, wherein the at least two target perfusion scan images correspond one-to-one with multiple critical phases; The second acquisition module is used to process the at least two target perfusion scan images through a perfusion parameter map generation model to obtain brain perfusion parameter images; The perfusion parameter map generation model is trained using multiple sample sets, each of which includes multiple sample scan images corresponding one-to-one with the multiple critical periods, and sample perfusion parameter images.

10. A medical imaging device, characterized in that, The medical imaging device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the method as described in any one of claims 1-8.

Images